Structures of two molluscan hemocyanin genes:

Significance for gene evolution

Bernhard Lieb*, Benjamin Altenhein*, Jurgen Markl*, Alexandra Vincent, Erin van Olden, Kensal E. van Holde,
and Karen I. Miller
*Institute of Zoology, Johannes Gutenberg University of Mainz, D-55099 Mainz, Germany; and Department of Biochemistry and Biophysics,
Agricultural and Life Sciences 2011, Oregon State University, Corvallis, OR 97331-7305
Contributed by Kensal E. van Holde, January 31, 2001

We present here the description of genes coding for molluscan

hemocyanins. Two distantly related mollusks, Haliotis tuberculata
and Octopus dofleini, were studied. The typical architecture of a
molluscan hemocyanin subunit, which is a string of seven or eight
globular functional units (FUs, designated a to h, about 50 kDa
each), is reflected by the gene organization: a series of eight
structurally related coding regions in Haliotis, corresponding to
FU-a to FU-h, with seven highly variable linker introns of 174 to
3,198 bp length (all in phase 1). In Octopus seven coding regions
(FU-a to FU-g) are found, separated by phase 1 introns varying in
length from 100 bp to 910 bp. Both genes exhibit typical signal
(export) sequences, and in both cases these are interrupted by an
additional intron. Each gene also contains an intron between signal
peptide and FU-a and in the 3 untranslated region. Of special
relevance for evolutionary considerations are introns interrupting
those regions that encode a discrete functional unit. We found that
five of the eight FUs in Haliotis each are encoded by a single exon,
whereas FU-f, FU-g, and FU-a are encoded by two, three and four
exons, respectively. Similarly, in Octopus four of the FUs each
correspond to an uninterrupted exon, whereas FU-b, FU-e, and FU-f
each contain a single intron. Although the positioning of the
introns between FUs is highly conserved in the two mollusks, the
introns within FUs show no relationship either in location nor
phase. It is proposed that the introns between FUs were generated
as the eight-unit polypeptide evolved from a monomeric precursor,
and that the internal introns have been added later. A hypothesis
for evolution of the ring-like quaternary structure of molluscan
hemocyanins is presented.

emocyanins are extracellular oxygen transport proteins

found in two animal phylaArthropoda and Mollusca.
Although the hemocyanins of both phyla use binuclear copper
sites for oxygen binding, they exhibit substantial differences in
tertiary and quaternary structure. Whereas arthropod hemocyanins function as multisubunit aggregates of protomers of about
75 kDa, molluscan hemocyanins are built from very much larger
polypeptide chains, ranging from 350 to 450 kDa in mass. Each
such molluscan polypeptide chain is comprised of seven or eight
different functional units (FUs), connected by short linker
peptides. Each FU carries a pair of copper atoms, which serve
to bind an oxygen molecule. Each copper is liganded by three
histidine residues, forming a copper A site and a copper B site
in each FU (14). Recently, we and our collaborators reported
the sequence of the polypeptide subunit of the cephalopod
mollusk Octopus dofleini (2), the 2.3- tertiary structure of the
C-terminal FU of this subunit (3), the polypeptide subunit
sequence of the gastropod mollusk Haliotis tuberculata (4), and
a 12- reconstruction of the quaternary structure of this hemocyanin (5). The polypeptide chain of Octopus hemocyanin
(OdH) was found to contain 2,896 aa residues, divided into seven
FUs, denoted OdH-a to OdH-g (from N to C terminus). The
polypeptide chain of Haliotis hemocyanin (HtH) (isoform 1)
consists of a corresponding set of FUs (termed HtH1-a to
HtH1-g), plus an additional C-terminal unit (HtH1-h) that
carries a specific tail extension of about 95 aa; in total, the

4546 4551 PNAS April 10, 2001 vol. 98 no. 8

protein subunit of HtH1 comprises 3,404 aa. The various FUs

exhibit considerable similarity (45% protein sequence identity) within as well as between the two mollusks; units in
corresponding positions have 55% identity.
On the other hand, comparison of either of these molluscan
protein sequences with those determined for subunits of arthropod hemocyanins (1, 6, 7) reveals only small similarity, mainly in
the region of one of the two copper binding sites, the B site. The
sequence surrounding the other site (the A site) resembles more
closely that found in tyrosinases (1, 2, 8, 9). A similar conclusion
is reached from comparison of the tertiary structure of the
Octopus FU OdH-g with the results from x-ray diffraction studies
of several arthropod hemocyanin subunits (3). Nonetheless the
overall geometry of the six histidine copper ligands that together
hold the two copper atoms is startlingly similar in the two phyla.
These observations raise important questions concerning the
evolution of these two classes of respiratory proteins and of the
phylogenetic relationships between arthropods and mollusks.
Current views would place these two phyla (once held to be
closely related) in entirely different superphylathe arthropods
as Ecdysozoans, the mollusks as Lophotrochozoans (see, for
example ref. 10). The limited similarities between the hemocyanins of mollusks and arthropods then would suggest their
independent evolution, following quite different paths, from a
(presumably monomeric) ancestral copper protein (1, 6, 8, 9, 11),
perhaps tyrosinase or phenoloxidase (12).
There is much more potential evolutionary information in
genes than in the corresponding cDNA. The sequence of one
arthropod hemocyanin gene has been published (13), but no
corresponding published data exists for the molluscan hemocyanins. Therefore, it seemed a logical next step to follow the
cDNA sequencing of HtH and OdH with determination of the
sequences of their genes. In sequencing the cDNA corresponding to the HtH, the two distinct isoforms HtH1 and HtH2
previously described at the protein level were found (14, 15);
counterparts termed KLH1 and KLH2 have been studied in
detail in another gastropod, the keyhole limpet Megathura
crenulata (16), and shown to exhibit differential regulation. The
hemocyanin isoforms HtH1 and HtH2 share only 65% sequence identity (4), and therefore their genes could be easily
distinguished. Here we present only the data on HtH1, as an
example of a gastropod hemocyanin gene sequence. Also in
Octopus dofleini, two variants of the hemocyanin sequence,
referred to as the A and G forms (OdHA and OdHG), were found
(17), but they are very similar (96%), in contrast to the
gastropod hemocyanin isoforms. In this report, we present only
Abbreviations: FU, functional unit; HtH, Haliotis tuberculata hemocyanin; OdH, Octopus
dofleini hemocyanin; UTR, untranslated region.
Data deposition: The sequences reported in this paper have been deposited in the GenBank
database [accession nos. AJ252741 (HtH1) and AF338426 (OdHG)].
To

whom reprint requests should be addressed. E-mail: Markl@mail.uni-mainz.de.

The publication costs of this article were defrayed in part by page charge payment. This
article must therefore be hereby marked advertisement in accordance with 18 U.S.C.
1734 solely to indicate this fact.

www.pnas.orgcgidoi10.1073pnas.071049998

the data for OdHG, as an example of a cephalopod hemocyanin

gene sequence.
Materials and Methods
Animals. The European abalone H. tuberculata (Archaeogastropoda) was a gift from the Syndicate Mixte d Equipment du
Littoral, Blainville sur Mer, France. The Pacific octopus, O.
dofleini, was obtained at the Oregon State University Marine
Science Center, Newport, OR.
Isolation and Analysis of Genomic DNA from Haliotis. Genomic DNA
was isolated from two mixed tissues (heart, mantle, 250 mg) of
a single adult abalone by using the Stratagene genomic-DNA
isolation kit according to the manufacturers instructions. PCR
was performed by using more than 50 different specific oligonucleotides (Genaxis, Spechbach, Germany) derived from the
cDNA coding for HtH1. For generating PCR fragments up to 2.5
kbp, a standard three-step PCR protocol with Taq polymerase
(GIBCO) was applied (94C for 20 s; 55C for 30 s; 72C for 120 s;
35 cycles). For larger fragments, we used the Expand-PCR
system (Roche Diagnostics) according to the manufacturers
instructions. PCR fragments were analyzed in standard agarose
gels in 1 TBE (89 mM Tris/chloride89 mM sodium borate2
mM sodium EDTA, pH 7.5), purified by the gel extraction kit
from Qiagen (Chatsworth, CA), and directly sequenced (Seqlab,
Go
ttingen, Germany), using the PCR oligonucleotides as primers. Alternatively, cloning in pGemT easy (Promega) was performed, and positive clones were subsequently sequenced with
standard andor gene-specific primers. The sequences obtained
were computer-analyzed as described (4).
Isolation and Analysis of Genomic DNA from Octopus. Genomic DNA
was isolated from O. dofleini tissues (brain, branchial gland, and
gonad) of an adult octopus by using the Qiagen DNEasy kit.
Fragments of the hemocyanin gene were obtained from the total
genomic DNA by using PCR (PTC-150 minicycler, MJ Research, Cambridge, MA). Primers for PCR were designed whenever possible from the cDNA sequences for this hemocyanin.
Oligonucleotide primers were synthesized by the Oregon State
University Central Services Laboratory. The DNA used was
isolated from an animal that had both A and G type hemocyanins. Because the sequence differences between these two forms
are small, care was taken to have redundant overlaps. Some of
the primers were A and G type-specific, but in other cases it was
necessary to clone the PCR products into the cloning vectors
pTrc His2-TOPO or pCR4-TOPO (Invitrogen) to obtain clones
that represented homogenous (A or G) products. PCR products
or sequences within vectors were sequenced in the Oregon State
University Central Services Laboratory. MACVECTOR (version
5.0.2) and ASSEMBLYLIGN (version 1.0.9b) (Oxford Molecular,
Oxford) were used to produce sequence contigs and alignments.
Lieb et al.

Results
The sequences of the HtH1 gene and the OdHG gene were
determined. Fig. 1 gives a schematic picture of the gene structure; for more details, see Fig. 5, which is published as supplemental material on the PNAS web site, www.pnas.org. Exon
lengths are given in Table 1. The complete nucleotide and
protein sequences of both genes can be found through GenBank.
Both coding sequences begin with a typical signal peptide. A
stop codon is observed near the 3 end of the terminal exon in
each gene (TAG in Haliotis and TAA in Octopus). As usually
observed, the sequence AATAAA also is found before the
poly(A) attachment site. As Fig. 1 shows, two types of introns can
be recognized in these genes. Introns in the linker region
between two functional units will be referred to as linker introns.
Introns within a functional unit will be called internal introns. All
but one of the splice junctions follow the GTAG rule for
Table 1. Size comparison of the different exons encoding HtH1
and OdHG
Haliotis

Octopus

Exon

bp

aa

Exon

s1
s2
s1s2
a1
a2
a3
a4
a1a4

32*
16
48
350
221
255
407
1233

411

b
c
d

1245
1242
1239

415
414
413

e
f1
f2
f1f2
g1
g2
g3
g1g3
h

1260
191
1060
1251
219
164
826
1209
1535

420

bp

15*

s1
s2
s1s2

36
28
64

417

a
b1
b2
b1b2
c
d
e1
e2
e1e2
f1
f2
f1f2

1233
463
782
1245
1242
1248
1019
241
1260
659
583
1242

403
511

1220

aa

21

411

415
414
416

420

414

407

Exon size is given in bp and aa; the total numbers per FU also are shown. For
sequence identities between the FUs in percent, see ref. 4. * indicates that the
signal sequence is still incomplete.
PNAS April 10, 2001 vol. 98 no. 8 4547

EVOLUTION

Fig. 1. Gene structures of HtH1 and OdHG. Exons encoding FUs (a h) are shown in different colors; introns (gray for linker introns, black for internal introns)
are shown to scale. s1 and s2 correspond to the two exons encoding each signal sequence. utr indicates the 3 UTR.

Fig. 2. Comparison of intron locations in molluscan hemocyanins and several

related proteins. Internal introns are shown by white arrows, linker introns by
gray arrows. The copper-binding sites (A and B), with their histidines (H) are
also schematically included. The -sandwich domain in molluscan FUs is indicated by the dotted region, with the hinge location marked. The unique tail
of FU-h is also indicated. Notation (aside from molluscan hemocyanin FUs):
AgP Anopheles gambia (insect) prophenoloxidase; NcT, Neurospora crassa
(fungus) tyrosinase; MmT, Mus musculus (mouse) tyrosinase, EcH-e, Eurypelma californicum (tarantula) hemocyanin subunit e.

donoracceptor sites (18); the exception is the intron in the

Octopus 3 untranslated region (UTR), which begins with AT.
Linker introns invariably start with a strictly conserved GT
sequence, and in most cases stop with a conserved CAG triplet.
Exceptions are intron HtH1(a4b), which is TAG, and introns
OdHG(cd) and OdHG(f2g), which are AAG. Details are
illustrated in Table 2, which is published as supplemental
material. All linker introns in both organisms show phase 1. In
both organisms the same is true for the intron between signal
peptide and FU-a and the introns found within the 3UTR. In
contrast, the nine internal introns (six in Haliotis and three in
Octopus) exist in all three phases.
Databank searches using only intron sequences showed no
significant homologies to transposable elements, ORFs for (regulatory) peptides or tRNA genes. The introns HtH1(s1s2) and
HtH1(g1g2) both contain a (GT)n-rich microsatellite. A prominent microsatellite region also is found within the internal
intron OdHG(e1e2). This microsatellite is unique in containing
quite long (ACAT)n, (GT)n, and (AT)n segments, with n as large
as 40 in the latter case.
Discussion
Comparison of the Two Molluscan Hemocyanin Genes: Exon-Intron
Structure. The structure of the HtH1 gene and the OdHG gene

are compared schematically in Fig. 1. Here we have distinguished

exons corresponding to different functional units by different
colors. In both cases, the most striking feature of the gene
organization is the presence of a linker intron between each pair
of FUs (see Figs. 2 and 3). Additionally, in both species, an intron
cuts the coding region of the signal peptide into two exons, a
second intron separates the signal peptide region from the first
4548 www.pnas.orgcgidoi10.1073pnas.071049998

Fig. 3. Sequence surrounding the linker introns. The protein linker regions
between the various FUs (designated by the respective exon pair) have been
arranged so as to (i) align conserved residues upstream from the linker intron
insertion sites (indicated by dotted lines) and (ii) align the (IV)RK consensus
downstream from the insertion site. Strongly conserved residues are marked
in black; there are many more of them in the -sheet region of each preceding
exon, to the left of the intron insertion site. The residues to the right of the site
constitute the protein linker sequence; there is little conservation here until
the (IV)RK motif, which marks the beginning of the first -helix in the next
exon. To make both alignments, it has been necessary to introduce gaps in
some of the linker sequences; their positions are not well defined. The
positions of two secondary structure elements as deduced from OdH-g (3) are
marked in light gray, and the positional conservation of a small amino acid is
marked in dark gray.

FU, and another intron inserts in the 3 UTR. In only three of

the eight FUs of HtH (HtH1-a, HtH1-f, and HtH1-g) and in
three of the seven FUs of OdH (OdHG-b, OdHG-e, and OdHG-f)
do we observe a division of the FU into two or more exons by
internal introns. The other nine FUs are each encoded by a
single, continuous exon. Because each FU of the protein contains about 400 aa, this means that many of the exons are
unusually large (see Table 1). Exons corresponding to FUs not
interrupted by internal introns average around 1,250 bp, and that
for HtH1-h is nearly 300 bp longer. On the other hand, most of
the introns are relatively short compared to those of higher
organisms (19).
The cephalopod hemocyanin gene especially exhibits a considerably higher exonintron ratio than do most eukaryotic
structural genes, with a consequence that the overall gene is not
extremely long (about 14.8 kbp from the start of the signal
peptide to the 3 UTR), with an ORF of 8,688 bases for the
secreted polypeptide, plus 63 bases for the signal peptide. The
gastropod hemocyanin gene is twice as large (28.6 kbp from the
signal peptide region to the 3 UTR), with a reading frame of
10,212 bases for the secreted polypeptide.
All of the linker introns in both genes have phase 1. Each of
these linker introns is inserted at a nearly equivalent point just
downstream from a conserved region at the C terminus of the
preceding FU, and upstream from a highly conserved sequence
[(VI)RK] in the following FU (see Fig. 3). This means that each
protein linker sequence is clearly defined and lies just downstream from the preceding linker intron. The internal introns
have all possible phases and are located in quite different
positions within each of their FU coding sequences (see Figs. 1
and 2, and discussion below). Among the six internal introns in
the HtH gene, three have phase 0, two have phase 2, and one has
phase 1; none of these introns is comparable to any of the three
internal introns (two of phase 0, one of phase 2) of the OdH
gene. Thus, the linker introns and internal introns present a
striking contrast. The regularity of the former can be explained
as a reflection of the genetic mechanism that created a multiunit
Lieb et al.

Comparison with Possibly Related Genes. Because molluscan and

arthropod hemocyanins exhibit certain sequence and structural

similarities (particularly in the region of the copper-binding
sites) it is of interest to compare the gene structures for OdH and
HtH with the only reported arthropod hemocyanin gene, that for
subunit e of the hemocyanin of the tarantula Eurypelma californicum (13). As Fig. 2 shows, there is very little similarity
between the genomic organization of any FU from Haliotis or
Lieb et al.

Octopus and that of the Eurypelma subunit. Whereas the 15

molluscan FUs are split by at most three introns (and in nine
cases by none), the gene for the arthropod subunit (which
corresponds in function to a single molluscan FU) contains no
less than eight introns and spans about 55 kbp, about four times
the length of the OdH gene and twice the length of the HtH gene,
which code for 5- to 6-fold more protein sequence, respectively.
These genetic differences reinforce the idea that arthropod and
molluscan hemocyanins are only distantly related. Indeed, recently Burmester (25) has argued by a convincing statistical
approach that arthropod and molluscan hemocyanin sequences
are not related at all. Molluscan hemocyanins do exhibit sequence similarity with tyrosinases, especially in regions surrounding both copper-binding sites (1, 2, 8, 9). However, comparison of the exon-intron structure of these two classes of
proteins indicates little, if any, similarity (Fig. 2). This is perhaps
not surprising, because comparison of the two molluscan genes
that are clearly related shows little similarity in internal intron
placement.
Evolutionary Implications. Considering all of the available data, it

seems likely that both tyrosinases and molluscan hemocyanins

evolved from a polypeptide chain that contained the copper A
and B regions (1, 2, 4, 8, 9). The highly regular placement of the
linker introns with respect to each preceding FU exon points to
a repetitive, simple process for generating the multifunctional
unit structures of molluscan hemocyanins from such a primordial unit. A possible scheme is shown in Fig. 4. Three rounds of
gene duplication and fusion using an identical splice site would
generate an eight-unit structure as seen in HtH. This is the kind
of mechanism that has been proposed for the evolution of certain
multiunit invertebrate hemoglobins (23, 2628). However, to
obtain the existing gene, another step would have been required;
the N-terminal signal sequence had to be added. Fungal tyrosinases (and presumably the putative precursor unit for the
hemocyanin gene as well) lack signal sequences; they are cytoplasmic proteins with enzymatic, not transport, functions. It is
hard to judge when the signal was added, for two extreme models
are possible. On the one hand, the signal sequence may have
been added only after the rounds of duplication generating the
eight-unit polypeptide chain were completed. Evidence from
comparison of FU sequences in both Octopus and Haliotis
indicates that this series of duplication was very rapid in evolutionary time (2, 4). Alternatively, the signal sequence could have
been added before gene duplications. In this model, the linker
protein sequences that lie between the FUs, downstream from
each linker intron, could represent the much-evolved relics of
what was originally the signal sequence of a single-unit progenitor. We tend to favor the latter view, for it provides a rationale
both for hemocyanin functional evolution and the development
of linker peptides. Perhaps the first use of a hemocyanin for
oxygen transport depended on the evolution of a signal sequence, permitting export of a monomeric oxygen-binding protein into the hemolymph. Gene duplication and fusion, building
larger proteins, then could allow higher concentrations of the
oxygen transporter in the blood, while maintaining low osmolality and low viscosity, and permitting high cooperativity (29).
The above scenario requires that the OdH has lost a preexisting FU, to give the number seven. From where was the unit
lost? Comparison of the cDNA sequence of the hemocyanin
from Haliotis with that from Octopus shows strong pairwise
similarity between FUs ag in the two organisms; this suggests
that it is the C-terminal unit (FU-h) that is missing in Octopus
(4). Moreover, in the phylogenetic tree derived from aligning
these sequences, HtH1-h forms a discrete branch that separates
from the other seven FU branches long before the gastropodcephalopod split (4). This observation (which is independent of
whether the unique tail extension of FU-h is taken into account
PNAS April 10, 2001 vol. 98 no. 8 4549

EVOLUTION

structure from a monomeric precursor (see below). These

introns presumably have remained in nearly the same position
and phase since the evolution of the multiunit structure, roughly
750 million years (4). In contrast, the internal introns appear to
be scattered, both in position and phase, in a random manner
within each of the hemocyanin genes, and between them. The
situation stands in stark contrast to that of the globin genes, in
which a few internal intron positions have been held over eons
(20). How can the apparent randomness in internal intron
position in hemocyanin genes be explained? Intron sliding from
one or a few initial positions can be ruled out; there is a paucity
of evidence for sliding in any genome, and it is very difficult to
explain how phase differences, as observed here, could be
generated (21, 22). In the absence of extreme sliding, we are left
with two alternatives: either (i) all of the internal introns
observed today are relics of a larger group of primordial introns
(introns early), or (ii) internal introns have been added (as well
as deleted?) over the evolutionary time since gastropods and
cephalopods arose (introns late). The first option seems
unlikely. If one assumes that the nine different internal introns
(six in Haliotis and three in Octopus) were all present in the
ancestral protomer gene (introns-early scenario), the original
eight-FU gene must have contained a total of 72 internal introns.
Consequently, it would be necessary to invoke a minimum of 69
independent intron losses in the OdH gene and of 66 independent losses in the HtH gene to produce the present distribution.
Moreover, in all nine cases of intron survival within this hypothetical hemocyanin gene, seven of the eight original copies of
each intron would have to have been wiped out, leaving no
common introns in the gastropod and cephalopod lines. Such a
scenario seems highly improbable. In contrast, if one follows the
intron-late scenario for the internal introns, only six and three
independent intron gains, perhaps accompanied by some losses
of earlier introns, would have been required for Haliotis and
Octopus, respectively. Thus, the genomic organization of these
molluscan hemocyanins appears to support the intron-late
model. In addition, it strongly suggests that: (i) The eight FU
genes arose by repeated gene duplication and fusion, possibly by
unequal crossing over (23), thereby producing the linker introns.
(ii) In the ca. 750 million years since that time the linker introns
have changed their sequence beyond recognition, but retained
their phase and nearly the same position. (iii) The internal
introns have been randomly inserted after the separation of
gastropods and cephalopods, ca. 520 million years ago.
Finally, we may ask whether or not the internal introns define
functional modules within a FU, as expected by some versions of
the intron-early model (24). As Fig. 2 shows, this is not the case.
According to the x-ray diffraction studies of OdH-g (3), there are
two well-defined domains within the molluscan functional unit,
the N-terminal, -helical core domain, which carries the oxygenbinding site, and a -sandwich domain rich in -sheets; the
domains are connected by a hinge. All but one of the introns
recognized in this work lie in the -helical domain, and none is
found in the hinge region between domains (Fig. 2; see also Fig.
6, which is published as supplemental material). Thus, there is
little evidence to relate existing internal introns to the stepwise
assembly of a modular structure. Indeed, if these introns have
been added after the evolution of a multi-FU structure (as we
argue above) there would be no reason to expect such evidence.

Fig. 4. Hypothesis for the evolution of the eight-FU molluscan hemocyanin subunit from a mono-FU precursor. Although our present results strongly suggest
that the eight-FU subunit evolved from three subsequent gene duplication and fusion events, a pedigree of the latter could not be constructed from the sequence
data so far. Thus, for the present scenario, recently revealed details of the quaternary structure (5) and general considerations of the evolution of extracellular
oxygen carriers (29, 30) have been taken into account. It is proposed that the dimeric repeating unit as well as the di-pentameric architecture of the extant
molluscan hemocyanin evolved before the eight-unit polypeptide, and that the latter grew from the C-terminal to the N-terminal FU and not vice versa. denotes
the signal peptide that probably gave rise to the different linker peptides.

or not) further supports the idea that the seven-unit hemocyanin

found in Octopus evolved from an eight-unit progenitor and not
the other way around.
The present study reveals no homology between the linker intron
sequences, which would allow one to calculate a pedigree, as in the
case of Artemia and Daphnia hemoglobin (27, 28). Thus, on the
basis of the available sequence data alone, the order of the evolution
of the eight hemocyanin FUs remains obscure, and further speculation requires information from other structural levels. A recent
high-resolution (12 ) three-dimensional reconstruction of the
HtH1 di-decamer revealed that the di-pentameric architecture of
the half molecule (the decamer) is mirrored by its internal collar,
a rather compact di-pentameric ring; this collar is composed of 10
copies of FU-h, arranged as five symmetrical dimers (5). In addition, the 12- reconstruction indicated an antiparallel dimer of
subunits to be the repeating unit, with an arc piece composed of
two FU-g as its stabilizing center. FU-g alone is able to stabilize the
repeating unit in Octopus, and moreover, it exhibits the most
conservative sequence (4), further indicating that it indeed plays a
crucial role in the formation of the dimeric repeating unit. Also
FU-f and FU-e participate in the dimeric core of the repeating unit,
whereas FU-d to FU-a are responsible for shaping the periphery of
the subunit dimer at its two ends (5). Our idea is that under the
selection pressure to form a larger respiratory protein (see above),
the molluscan hemocyanin started from a FU homodimer, which
later oligomerized into a di-pentameric ring; comparable protein
structures have been observed in invertebrate hemoglobins (30).
4550 www.pnas.orgcgidoi10.1073pnas.071049998

Then the core and finally the periphery of the extant decameric
cylinder evolved (Fig. 4; see also Fig. 7, which is published as
supplemental material). The intriguing aspect of this idea is that it
proposes that the dimeric repeating unit as well as the dipentameric architecture of molluscan hemocyanin pre-existed the
evolution of the eight-unit polypeptide. In this context it would
seem unlikely that this process started from FU-a, whereas the
opposite possibility, an origin from FU-h, is supported by structural
considerations. We propose that an FU-h precursor equipped with
a secretion signal (h) and able to dimerize after secretion (to form
a h-h homodimer) became able to oligomerize into a di-pentameric
( decameric) ring (Fig. 4). Next, the ancient single-unit polypeptide grew by three subsequent steps of gene duplication and fusion,
producing a stepwise transformation of the simple ring into the
highly complex molluscan hemocyanin cylinder (Fig. 4). Specifically, we propose: (i) A first gene duplication and fusion yielded
hh, which then evolved to gh, both still forming a pair of
homodimers within the repeating unit (5). (ii) A subsequent gene
duplication, fusion, and further evolution produced efgh,
which formed a larger homodimer, further consolidating the core
of the repeating unit (Fig. 4). (iii) A third duplication, fusion, and
evolution gave abcdefgh, with two copies of the
FU-a to FU-d sequence in each dimer of subunits furnishing the
two peripheral enlargements of the core (Fig. 4). The tail extension
of FU-h, which is not reflected in any of the other FUs (see Fig. 2),
must have occurred at a later date. As discussed above, the copies
of the primordial signal peptide () subsequently transformed, and
independently evolved, into the different linker peptides.
Lieb et al.

acknowledges the support of this research by National Science Foundation Grant MCB-9805570 to K.I.M. The work of the German group has
been financially supported by the biosyn company, Fellbach, and the
Deutsche Forschungsgemeinschaft (Grant Ma 8434-3 to J.M.).

1. van Holde, K. E. & Miller, K. I. (1995) Adv. Protein Chem. 47, 181.
2. Miller, K. I., Cuff, M. E., Lang, W. F., Varga-Weisz, P., Field K. & van Holde,
K. E. (1998) J. Mol. Biol. 278, 827842.
3. Cuff, M. E., Miller, K. I., van Holde, K. E. & Hendrickson, W. A. (1998) J. Mol.
Biol. 278, 855870.
4. Lieb, B., Altenhein, B. & Markl, J. (2000) J. Biol. Chem. 275, 56755681.
5. Meissner, U., Dube, P., Harris, J. R., Stark, H. & Markl, J. (2000) J. Mol. Biol.
298, 2134.
6. Markl, J. & Decker, H. (1992) Adv. Comp. Env. Physiol. 13, 325376.
7. Voit, R., Feldmaier-Fuchs, G., Schweikardt, T., Decker, H. & Burmester, T.
(2000) J. Biol. Chem. 275, 3933939344.
8. Lerch, K., Huber, M., Scheider, M.-J., Drexel, R. & Linzen, B. (1986) J. Inorg.
Biochem. 26, 213217.
9. Durstewitz, G. & Terwilliger, N. B. (1997) Mol. Biol. Evol. 14, 266276.
10. Conway Morris, S. (1998) Curr. Opin. Genet. Dev. 8, 662667.
11. van Holde, K. E. (1998) Zoology 100, 287297.
12. Decker, H. & Terwilliger, N. B. (2000) J. Exp. Biol. 203, 17771782.
13. Voll, W. & Voit, R. (1990) Proc. Natl. Acad. Sci. USA 87, 53125316.
14. Keller, H., Lieb, B., Altenhein, B., Gebauer, W., Richter, S., Stricker, S. &
Markl, J. (1999) Eur. J. Biochem. 264, 2738.
15. Lieb, B., Altenhein, B., Lehnert, R., Gebauer, W. & Markl, J. (1999) Eur.
J. Biochem. 265, 134144.

16. Harris, J. R. & Markl, J. (1999) Micron 30, 597623.

17. Lang, W. H. (1988) Biochemistry 27, 72677282.
18. Long, M., de Souza, S. J., Rosenberger, C. & Gilbert, W. (1998) Proc. Natl.
Acad. Sci. USA 95, 219223.
19. Hawkins, J. D. (1998) Nucleic Acids Res. 16, 98939905.
20. Jellie, A. M., Tate, W. P. & Trotman, C. N. A. (1996) J. Mol. Evol. 42, 641647.
21. Li, W.-H. (1997) Molecular Evolution (Sinauer Associates, Sunderland, MA),
pp. 379418.
22. Stoltzfus, A., Logsdon, J. M., Jr., Palmer, J. D. & Doolittle, F. W. (1997) Proc.
Natl. Acad. Sci. USA 94, 1073910744.
23. Naito, Y., Riggs, C., Vandergon, T. L. & Riggs, A. F. (1991) Proc. Natl. Acad.
Sci. USA 88, 66726676.
24. Gilbert, W., Marchionni, M. & McKnight, G. (1986) Cell 46, 151153.
25. Burmester, T. (2001) Mol. Biol. Evol. 18, 184195.
26. Moens, L., Vanfleteren, J., De Baere, I., Jellie, A. M., Tate, W. & Trotman,
C. N. A. (1992) FEBS Lett. 312, 105109.
27. Matthews, C. M. & Trotmann, C. N. A. (1998) J. Mol. Evol. 47, 763771.
28. Hebert, P. D. N., Um, Y. M., Prokopowitch, C. D. & Taylor, D. J. (1999) J. Mol.
Evol. 49, 769779.
29. Mangum, C. P. (1998) Am. Zool. 38, 113.
30. Ilan, E., David, M. M. & Daniel, E. (1981) Biochemistry 20, 61906194.

EVOLUTION

The American group is especially grateful to colleague Andrew Karplus

for stimulating discussion. The careful and diligent assistance of Kristin
Rorrer as well as the staff in the Oregon State University Central
Services Laboratory is gratefully acknowledged. The American group

Lieb et al.

PNAS April 10, 2001 vol. 98 no. 8 4551